Académique Documents
Professionnel Documents
Culture Documents
countless hours spent laying out the incredibly small traces that made the autopilot board
as small and light as it is. I also want to acknowledge Derek Kingston for his work in
tackling the elusive attitude estimation problem. Josh Hintz and Dave Hubbard deserve
credit on three accounts: First, for sticking with the project in the early days when
working on the autopilot was not fun; Second, for spending long hours in the sun testing
doomed autopilot algorithms while everyone in the lab thought we were out sunbathing at
the park; Third, for developing the venerable Virtual Cockpit ground station. In addition,
I would like to thank Kendall Fowkes for the hours spent designing and building
airplanes only to have them broken into a million pieces. I would also like to thank
Carolyn Cornaby for having the courage to edit and proofread a 240 page thesis. Finally,
I would like to thank the following people for their many contributions: Steve Griffiths,
Josh Redding, Morgan Quigley, Deryl Snyder, Mark Denny, and Jen Wilds.
CONTENTS
Acknowledgments
vii
List of Figures
xiii
List of Tables
xvii
Introduction
Flight Control
2.1
2.1.1
2.1.2
2.1.3
2.1.4
2.1.5
2.2
2.2.1
2.2.2
Longitudinal Control..................................................................................19
vii
2.3
2.3.1
2.3.2
Autopilot Hardware
29
3.1
3.2
3.2.1
3.2.3
Electric Conversion....................................................................................36
3.3
Avionics .........................................................................................................37
3.3.1
3.3.2
I/O ..............................................................................................................47
3.3.3
3.3.4
3.3.5
Bypass Circuit............................................................................................53
3.3.6
Payload.......................................................................................................56
3.3
4
Autopilot Software
4.1
61
4.1.1
4.1.2
Main Loop..................................................................................................66
4.1.3
Costates ......................................................................................................68
4.2
4.2.1
viii
4.2.2
4.3
4.3.1
4.3.2
Estimating Heading....................................................................................78
4.4
Sensor Calibration..........................................................................................79
4.4.1
Autopilot Sensors.......................................................................................80
4.4.2
4.5
4.6
4.6.1
4.6.2
4.7
High-Level Control........................................................................................96
4.7.1
4.7.2
PIC Fail-Safe..............................................................................................98
4.8
4.9
Autopilot Performance.................................................................................100
105
5.1.1
Airborne Components..............................................................................106
5.1.2
5.2
5.2.1
5.2.2
ix
5.2.3
5.2.3
5.3
5.3.1
5.3.2
5.3.3
5.3.4
5.4
5.4.1
5.4.2
5.4.3
5.4.4
5.4.5
5.4.6
5.4.7
5.5
First Flight....................................................................................................200
5.6
5.6.1
5.6.2
5.6.3
5.7
Pre-Flight .....................................................................................................207
5.7.1
Power-Up .................................................................................................207
5.7.2
Control Surfaces.......................................................................................208
5.7.3
GPS ..........................................................................................................208
5.7.4
Sensors .....................................................................................................209
5.7.5
Waypoint Programming...........................................................................210
5.7.6
5.7.7
5.7.8
Launch......................................................................................................211
5.8
5.9
Landing ........................................................................................................214
Appendix A
217
Appendix B
219
Appendix C
231
Appendix D
233
BIBLIOGRAPHY
235
xi
xii
LIST OF FIGURES
Figure 3.4: Completed Zagi THL flight platform with electric conversion. .................... 37
Figure 3.5: Avionics hardware diagram............................................................................ 38
Figure 3.6: Autopilot main board block diagram.............................................................. 39
Figure 3.7: In-house IMU hardware. ................................................................................ 45
Figure 3.8: Sensor Sampling Diagram.............................................................................. 46
Figure 3.9: Autopilot board showing interface ports. ....................................................... 48
Figure 3.10: Furuno GH-80 GPS installed. ...................................................................... 49
Figure3.11: Aerocomm AC4490-500-M3 digital modem installed in a UAV................. 51
Figure 3.12: 900 MHz digital communications dipole antenna installed in a UAV. ....... 52
Figure 3.13: Photo of the completed Bypass circuit. ........................................................ 54
Figure3.14: Bypass circuit diagram. ................................................................................. 55
Figure 3.15: Bypass circuit connected to the autopilot, servos, and RC receiver............. 56
Figure 3.16: Video payload installed in UAV. ................................................................ 57
Figure 3.17: Ground station hardware. ............................................................................. 59
Figure 4.1: Autopilot software high level diagram.......................................................... 62
Figure 4.2: Pre-main software flow. ................................................................................ 63
Figure 4.3: Global variables and structures. .................................................................... 64
Figure 4.4: Main loop flow and autopilot costates. ......................................................... 67
Figure 4.5: Sensor processing flow.................................................................................. 70
Figure 4.6: Attitude estimation flow................................................................................ 76
Figure 4.7: Heading estimation flow. .............................................................................. 79
Figure 4.8: Basic PID loop structure as implemented in the autopilot source code. ....... 83
Figure 4.9: Aileron control loop as implemented in the autopilot source code............... 84
xiv
xv
xvi
LIST OF TABLES
xvii
xviii
CHAPTER 1
INTRODUCTION
Unmanned Aerial Vehicles (UAVs) have proven their usefulness in military
applications in recent years. Large UAVs such as the General Dynamics Predator (27
foot wingspan) [1] have become an integral part of the U.S. arsenal. Since 1994, the
Predator has logged more than 20,000 flight hours, executing surveillance and tactical
missions in virtually every part of the world. The military is also making use of smaller
UAVs, known as mini UAVs. The AeroVironment Pointer Mini-UAV was one of the
first Mini UAVs deployed [2]. It has been used all over the world by small military units
as a short range video reconnaissance platform. It has a cost of $88,000 [3], relatively
low in comparison with the Predators price tag of $50 million. The Pointer has proven
itself robust and reliable, but it does have disadvantages. This UAV is designed to be
man-packable but the UAV is heavy and cumbersome for one person to carry. Its 9foot wingspan also makes it easily visible to the enemy at the low altitudes it flies.
Hence, the armed forces have made a significant push to develop a replacement for
Pointer. This effort has lead to the Dragon Eye Mini UAV. The Dragon Eye is smaller
and lighter than Pointer. However, Dragon Eye is still large enough (4 foot wing span)
[4] to be visible to the enemy and requires a 2-man crew to transport and operate. There
is room for improvement in the areas of UAV size and portability.
Recently, interest has increased in the relatively undeveloped area of UAVs with
wingspans of less than 2 feet. Military personnel, encouraged by the successes of Pointer
and Dragon Eye, desire to get the same close area surveillance capabilities into smaller,
lighter packages. These Micro UAVs have several advantages over Mini UAVs. Their
small size makes them easier to carry. It is possible for a Micro UAV to be transported
and operated by one individual, as opposed to the two- or three- man crews that are
currently required to operate the Mini UAVS. The size of Micro UAVs also allows them
to be packed in and deployed from small containers. For one application, this would
allow several Micro UAVs to be deployed from a Mini or Large UAV to gather closerange surveillance data. Micro UAVs are also less expensive than their larger
counterparts. They are smaller and require less material to build. This cost saving gives
them potential to be a disposable asset; Micro UAVs could be flown in unfriendly areas
where the risk of damage is high. It also frees the operator from the task of ensuring the
airplane is recovered, allowing more effort to be spent on primary mission goals.
In addition to military applications, there are many potential civilian
applications for Micro UAVs [5]. The size and cost advantages that make them
appealing to the military also make them attractive for use in the private sector.
However, the use of UAVs in non-military applications is currently limited for reasons of
cost, safety, and special requirements. The prohibitive cost of Large and Mini UAVs
makes it difficult for individuals and private organizations to purchase and own. The
current generation of Mini and Large UAVs also pose a potential danger to the civilian
population because of their weight. Micro UAVs are much safer in this aspect, as they
can weigh less than 1 kilogram. In addition to high cost and physical danger associated
with currently available UAVs, the challenges of using these aircraft in the private sector
include significant training of the operator. Large UAVs also require large runways to
operate.
Micro UAVs have the potential to overcome many of the limitations of Mini and
Large UAVs. For this reason there is great interest in adopting Micro UAVs for a variety
of applications. These include government applications, such as fire fighting and law
enforcement. The U.S. spends millions of dollars each year patrolling borders with
conventional aircraft [6]. If border agents were equipped with low-cost Micro UAVs,
thousands of dollars could be saved, as UAVs cost much less then similarly equipped
full-size aircraft. In the private sector, there also exists a range of surveillance
applications for UAVS. One example is use by the media. Newscasters spend millions
of dollars on helicopters to cover breaking news stories. Micro UAVs could be carried
by news crews and launched over a breaking story to gather aerial footage immediately.
The Micro UAVs would also pose less risk to the civilian population than low-flying fullsized helicopters. It is likely that as Micro UAVs become available, they will be
employed in many more applications to cut cost and improve coverage.
For Micro UAVs to be successful militarily and commercially, they must be
low-cost, easy to operate, and small. Currently, there are no fully autonomous
operational UAVs available with wingspans of less than 2 feet. One of the main
roadblocks in developing Micro UAVs is the size and weight of the autopilot system. For
an aircraft with a wingspan of less than 2 feet to carry an autopilot and useful payload,
the autopilot cannot weigh more than a few ounces. It must also be small enough to fit in